Is the second nic intended for bonding or other application? Is there an
IP address it should have?
Discovery should have disabled the second nic or assigned the intended
address. To manually do so, nodech m2 mac.mac='6c:ae:8b:20:d5:32|
6c:ae:8b:20:d5:33!*NOIP*' then makedhcp.
FYI, you could also have left netboot=xnba and the 'right' behavior would
have been observed (the firmware would never actually attempt the second
nic).
From: Lohit Valleru <loh...@gwmail.gwu.edu>
To: xCAT Users Mailing list <xcat-user@lists.sourceforge.net>
Date: 06/17/2014 01:57 PM
Subject: [xcat-user] Regarding post-discovery issue with dual ethernet
interfaces
Hello All,
xCAT server in one of our cluster environments is running with version :
2.7.4
This is a prebuilt xCAT cluster from IBM, and i really do not want to
update this test environment as of now.
Issue : ( Node constantly goes into discovery genesis kernel and does not
install OSimage )
I have followed the following steps:
nodeset m2 osimage=centos6-x86_64-install-bluefin-mgmt
rsetboot m2 net
rpower m2 boot
However, the node does not recognize that it has to install the above
OSimage. Instead, it goes into the discovery mode, everytime.
I had verified the tftpboot files, and also the dhcpd leases values. The
leases file clearly mentions that m2 node mac address (6c:ae:8b:20:d5:32 )
should load pxelinux.0 file. I do not find an issue with it.
After some analysis, i realized that the node is actually trying to dhcp
from "both" of the ethernet interfaces that are present on the server with
mac addresses:
6c:ae:8b:20:d5:32 and 6c:ae:8b:20:d5:33
6c:ae:8b:20:d5:32 has the node discovered with its definition.
However 6c:ae:8b:20:d5:33 does not have a node defined with it, which is
leading to node discovery and the node goes into genesis kernel with ip
address from the dynamic range.
Object name: m2
arch=x86_64
bmc=172.29.102.2
bmcpassword=PASSW0RD
bmcport=0
bmcusername=USERID
chain=runcmd=osimage=centos6-x86_64-install-bluefin-mgmt
currchain=boot
currstate=install centos6-x86_64-bluefin-mgmt
groups=mgmt
initrd=xcat/centos6/x86_64/initrd.img
installnic=mac
kcmdline=quiet repo=http://!myipfn!/install/centos6/x86_64/
ks=http://!myipfn!/install/autoinst/m2 ksdevice=6c:ae:8b:20:d5:32 cmdline
console=tty0 console=ttyS0,115200
kernel=xcat/centos6/x86_64/vmlinuz
mac=6c:ae:8b:20:d5:32
mgt=ipmi
mtm=7915AC1
netboot=pxe
ondiscover=nodediscover
os=centos6
postbootscripts=otherpkgs
postscripts=syslog,remoteshell,syncfiles
profile=bluefin-mgmt
provmethod=centos6-x86_64-install-bluefin-mgmt
serial=KQ0C1B6
serialport=0
serialspeed=115200
status=powering-off
statustime=06-17-2014 13:34:22
supportedarchs=x86,x86_64
switch=switch2
switchport=12
Is there anyway, that we could prevent this from happening.
Is it possible to use just one interface for tftp and dhcp look up.
It is also possible that, i do not understand the issue correctly. Please
advice.
Thank you,
Lohit
------------------------------------------------------------------------------
HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
Find What Matters Most in Your Big Data with HPCC Systems
Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
Leverages Graph Analysis for Fast Processing & Easy Data Exploration
http://p.sf.net/sfu/hpccsystems
_______________________________________________
xCAT-user mailing list
xCAT-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xcat-user
------------------------------------------------------------------------------
HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
Find What Matters Most in Your Big Data with HPCC Systems
Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
Leverages Graph Analysis for Fast Processing & Easy Data Exploration
http://p.sf.net/sfu/hpccsystems
_______________________________________________
xCAT-user mailing list
xCAT-user@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/xcat-user
